This repo contains the implementation of a simple INR that trains on a single image.
All the code and outputs are present in the notebook present in this repo.
Certain packages may need to be installed as per requirements.
The notebook may need require certain packages to be installed.
It is highly recommended to run on a system that has CUDA support.
For the sin activation function used in the code.
For Postional Encoding, the Paper Fourier Features Let Networks Learn High Frequency Functions in Low Dimensional Domains and the code provided by the authors publically inspired the implementation of the same above.